In this article, an online state of charge (SoC) estimation framework is proposed, designed and implemented using the system identification approaches. The techniques are composed of cross combination between two modified nonlinear optimisation algorithms (modified Genetic Algorithm and modified Levenberg Marquardt) adapted for battery cell parameter estimation. Subsequently a linear recursive Kalman filter is employed to estimate the state parameters of the battery cell. Moreover, a newly statistical approach is developed to encounter hysteresis phenomena within the cell. The prerequisite for the SoC determination in the electrical vehicle (EV) is challenging as the battery can be composed of hundreds of cells while the load current changes dramatically inside the cells and the required elapsed time for SoC determination should be as short as possible to extend the expected lifetime of the battery pack. Thus, the accurate estimation of the SoC of the cells in a battery pack is one of the key factors for using them effectively. The framework is found to be robust, optimal and implementable in time constrained environment with acceptable accuracy. (C) 2013 Elsevier B.V. All rights reserved.